Comparison of two assays for measuring plasma concentrations of paracetamolBMJ 1997; 315 doi: https://doi.org/10.1136/bmj.315.7114.991 (Published 18 October 1997) Cite this as: BMJ 1997;315:991
- C V Egleston, senior registrara,
- C Browning, senior house officerb,
- I Hamdi, senior registrarc,
- G Campbell-Hewson, registrar,
- S M Robinson, consultantb
- a Department of Accident and Emergency Medicine, Norfolk and Norwich Hospital, Norwich NR1 3SR
- b Department of Accident and Emergency Medicine, Addenbrooke's NHS Trust, Cambridge CB2 2QQ
- c Department of Clinical Biochemistry, Addenbrooke's NHS Trust
- Correspondence to: Dr Robinson
- Accepted 11 March 1997
An overdose of paracetamol can cause fatal hepatic necrosis, but acetylcysteine is an effective antidote if given early.1 An estimation of plasma concentrations of paracetamol is required if a patient is suspected of having taken a toxic dose. We compared the AcetaSite test card and Stat-Site reflectance meter (GDS Diagnostics, Elkhart, IN), a bedside method of determining paracetamol concentrations that takes 2 minutes, with the Quantase assay (Porton Products, Maidenhead), an established laboratory method.
Patients, methods, and results
Altogether 192 patients were recruited into the study; of these, 92 had taken paracetamol. Patients who had not ingested paracetamol but required venepuncture for other reasons made up the control group. A blood sample was obtained 4 hours after suspected overdose or on arrival if 4 hours had already passed. A drop of this blood was used to measure paracetamol concentration with the AcetaSite card and Stat-Site meter, and the remainder was sent to the laboratory. Medical and nursing staff were trained to use the equipment before the study started.
A statistical method first described by Bland and Altman2 was used to assess agreement between the laboratory standard and the AcetaSite test. This method plots the difference between results obtained by a standard and a new test against the mean of these measurements. Logarithmic transformation of the data was performed because the data were not normally distributed.
The 1 uses untransformed data to show the relationship between paracetamol concentrations obtained with the laboratory test and those obtained with the AcetaSite test. The limits of agreement are 0.16 and 5.04—that is, in 95% of cases the AcetaSite result was between 0.16 and 5.04 times the laboratory result. The results for five patients fell outside these limits. There were six false negative results but no false positive results. The median time for a result to be available from the laboratory was 28.5 minutes (range 3–166 minutes).
The wide limits of agreement in this study suggest that there are considerable discrepancies between the two methods in assessing plasma concentrations of paracetamol. Differences between the AcetaSite and laboratory results were spread randomly between high and low concentrations of paracetamol. Treatment with acetylcysteine would have been withheld from three out of the five patients whose results were outside the limits of agreement, had the decision to treat been based only on the result from the AcetaSite test. This might have resulted in a poor outcome.
Impressive performance characteristics are reported in the datasheet for the AcetaSite test and Stat—Site reflectance meter when compared with the GDS enzymatic liquid reagent (r>=0.970) and the TDX (Abbott) liquid reagent (r>=0.983). For the data sheet, accuracy was assessed using whole blood, plasma, and serum samples with known concentrations of paracetamol and a small number of clinical samples (n=42). The methodology may partially explain the discrepancy between the results found in our study and those found in preclinical testing; correlation does not assess the degree of agreement but rather the relationship between the two tests.2 Also, the use of samples with known concentrations in a laboratory environment may not accurately replicate analysis of samples obtained in a clinical setting.
Operator error may explain why some of the results from the AcetaSite test bear little relation to the results found with the laboratory tests. Although the majority of department staff attended two training sessions, difficulties in using the Stat-Site meter were reported by some inexperienced operators. The production of a simple algorithm for using the card and meter reduced the number of difficulties reported.
The rapidity with which the AcetaSite results were available could have been advantageous if the results had been in agreement with the laboratory results. This study found that the AcetaSite test should not replace the standard Quantase assay.
We wish to thank Dr C Palmer for his statistical advice and Miss C Reid for her help in data acquisition.
Funding: The equipment and technical support required for this study were provided by Cambridge Life Sciences.
Conflict of interest: None.